Real-time estimation of secondary crash likelihood on freeways using high-resolution loop detector data

This study aimed to develop a secondary crash risk prediction model on freeways using real-time traffic flow data. The crash and traffic data were collected on the I-880 freeway for five years in California, United States. The secondary crashes were identified by a method based on speed contour plot. The random effect logit model was used to link the probability of secondary crashes with the real-time traffic flow conditions, primary crash characteristics, environmental conditions, and geometric characteristics. The results showed that real-time traffic variables significantly affect the likelihood of secondary crashes. These traffic variables include the traffic volume, average speed, standard deviation of detector occupancy, and volume difference between adjacent lanes. In addition, the primary crash characteristics, environmental conditions and geometric characteristics also significantly affect the risks of secondary crashes. The model evaluation results showed that the predictive performance of the developed model was deemed satisfactory. The inclusion of traffic flow variables and random effect increases prediction accuracy by 16.6% and 7.7%, respectively. These results have the potential to be used in advanced traffic management systems to develop proactive traffic control strategies to prevent the occurrences of secondary crashes on freeways.